Exercise interventions in healthy older adults with sarcopenia: A systematic review and meta‐analysis
Abstract
Objective
To systematically assess the effects of exercise interventions on body composition and functional outcomes in older adults with sarcopenia.
Methods
PubMed/Medline, Embase and Cochrane Library were searched from 2006 to 2017 for exercise randomised controlled trials and controlled clinical trials in adults 60 years and older with sarcopenia. Preferred Reporting Items for Systematic Review and Meta‐Analysis protocol (PRISMA‐P) and Physiotherapy Evidence Database (PEDro) scale assessed internal validity. Meta‐analysis and sensitivity analysis were performed.
Results
Searches retrieved 1512 titles. Thirty‐two full texts were evaluated, and six trials were included. Methodological quality was 5.5 (0–10). Meta‐analysis revealed that knee‐extension strength (P ≤ 0.01), timed up and go (P < 0.0001), appendicular muscle mass (P = 0.04) and leg muscle mass (P = 0.04) significantly improved in response to exercise interventions.
Conclusions
Exercise interventions significantly improved strength, balance and muscle mass. However, the number of trials was small and the training effect was inconsistent due to heterogeneity in exercise mode, duration and intensity. Lack of detailed description makes it impossible to reflect on the progressive resistance training approaches used. More research is needed to confirm these findings.
Policy Impact
The burden of sarcopenia is extensive, particularly given our ageing population thus prevention and effective interventions are important. The cost‐effectiveness of interventions is also important. Currently there are no cost‐effectiveness exercise intervention trials that focus on sarcopenic older adults. However, in previous studies with older populations, exercise interventions were shown to be more cost‐effective than dietary interventions.
Practice Impact
Some exercise interventions have the potential to reverse or slow the sarcopenic process. However, the existing evidence is based on populations of differing ages. The inconsistent findings limit our understanding. There is currently insufficient evidence to enable definitive exercise intervention recommendations to be made.
Introduction
Sarcopenia is a well‐known geriatric syndrome and is recognised worldwide 1. The European Working Group on Sarcopenia in Older People (EWGSOP) is one of several groups that has provided a working definition of sarcopenia as ‘a syndrome characterised by progressive and generalised loss of skeletal muscle mass and strength with a risk of adverse outcomes such as physical disability, poor quality of life and death’ [2], which occurs from approximately 40 years of age onwards 3. This loss of muscle mass has been estimated at about 8% per 10 years until the age of 70 3. After reaching age 70, this loss increases to 15% every decade 3. The loss can eventually cause a 50% decrease in muscle mass by the age of 80 years 4. A systematic review by the EWGSOP on the prevalence of sarcopenia reported a prevalence of 1–29% in community‐dwelling older people, with even higher estimated prevalence in long‐term and acute care settings 5. Scott et al. 6 reported a prevalence of sarcopenia of 37% in community‐dwelling middle‐aged adults. Sarcopenia has also been classified into three progressive phases: (i) presarcopenia, which is a decrease in muscle mass, but not in strength and function; (ii) sarcopenia which is characterised by a decrease in muscle mass and muscle strength or function; and (iii) severe sarcopenia with significantly decreased muscle mass, strength and function 2.
Sarcopenia research has steadily increased over the past two decades 7, and much of this research is about the health consequences associated with sarcopenia, which include frailty 8, osteoporosis 9, low health‐related quality of life 9, obesity 10, dementia 11 and type 2 diabetes 12, as well as heart failure, chronic obstructive pulmonary disease and renal failure 8. Sarcopenia also predicts future mortality in middle‐aged and older adults 9, 12. One of the risk factors for sarcopenia is limited physical activity, which is common among older adults 13, 14. To mitigate the risk of sarcopenia, different exercise interventions have been tested in healthy older people without diagnosed sarcopenia. Resistance training is a commonly used exercise intervention and has shown positive effects on both muscle strength and physical function 15. Interventions to counteract physical inactivity among older adults are often part of government health strategies 16. Most interventions are aimed at preventing or slowing age‐related loss of function and muscle prior to a diagnosis of sarcopenia. However, evidence around exercise interventions, specifically targeted to older adults with sarcopenia, is lacking.
Therefore, the aim of this review was to systematically assess the effects of exercise interventions on body composition, muscle strength and functional outcomes in older adults with sarcopenia.
Methods
Three different electronic databases, PubMed/Medline, Embase and Cochrane Library, were searched for eligible articles from 2006 to 25 March 2017. The following MeSH terms and keywords were used either singularly or in combination: sarcopenia, aged, elderly, older adults, physical therapy, exercise, training, muscle strength and muscle mass. The complete search strategy is shown in Appendix S1 (Supporting information). Articles were eligible for inclusion if they were: (i) a full‐text article; (ii) published in English or Dutch; (iii) published in a peer‐reviewed journal; (iv) described as a controlled exercise‐based intervention trial; and (v) conducted in people 60 years of age and older with sarcopenia. Sarcopenia was classified according to the EWGSOP definition. Studies in other patient groups were excluded. Studies with participant groups in which all participants had a specific comorbidity (e.g. all participants had obesity); studies without a control group; studies not conducted on human beings; studies where the objective was the evaluation of a diagnostic instrument; and case reports, abstracts, editorials, opinions and dissertations were also excluded. Studies published before 2006 were excluded because this review aimed to focus on the latest and newest insights, which are fully applicable to the current health‐care system. The Preferred Reporting Items for Systematic Review and Meta‐Analysis protocol (PRISMA‐P) for systematic reviews was used as a guide to maximise the internal validity 17.
Study selection
Review of articles to determine eligibility was carried out by two reviewers (LV and WH). When titles and abstracts implied that an article was potentially eligible for inclusion, a full‐text report was reviewed. Additionally, reference tracking was performed on the included articles (flow diagram; Figure 1). Disagreements were resolved by consensus.

Quality assessment
Two reviewers independently assessed the methodological quality of the included articles (LV and WH) using the Physiotherapy Evidence Database (PEDro) scale for randomised controlled trials (RCTs) and controlled clinical trials. The PEDro scale is a domain‐based evaluation where a critical assessment can be made separately for 11 different domains such as allocation and blinding. This scale has undergone substantial psychometric validation 18. For the full PEDro scale, see Appendix S2 (Supporting information). Disagreements were resolved by consensus. The inter‐rater reliability was calculated by a Cohen's kappa where <0 is less than chance agreement, 0.01–0.20 slight agreement, 0.21–0.40 fair agreement, 0.41–0.60 moderate agreement, 0.61–0.80 substantial agreement and 0.81–0.99 almost perfect agreement 19.
The overall quality of evidence was summarised using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) system [20]. There are four levels in the GRADE system. It ranges from high‐quality evidence (further research is very unlikely to change the estimate or effect) to very low‐quality evidence (any estimate of effect is very uncertain). The quality of evidence is rated by downgrading the high‐quality level if one of the following criteria is present: limitations in study design and/or execution; inconsistency of results; generalisability of population and outcomes; and the reported odds ratio/relative risk/hazard ratio for comparison 19, 20.
Data extraction
One reviewer (LV) extracted the data using a standard extraction form. Data extracted from the included articles include the following: (i) authors, publication year, setting and study participants; (ii) number of participants; (iii) study design; (iv) participants exercise group; (v) control group intervention; and (vi) outcome measures. If data were missing or additional information was required, multiple attempts were made to contact the corresponding authors, to obtain the required information. Data extraction was checked by the second author (WH). Data on the exercise intervention were extracted based on the Consensus on Exercise Reporting Template (CERT) 21. For the original CERT checklist, see Appendix S3 (Supporting information).
Outcome measures
Two different types of outcome measure were assessed in this systematic review: functional outcome measures, such as grip strength and gait speed; and body composition outcome measures, such as muscle mass and body fat.
Statistical analysis
Descriptive statistics described the extracted data. Meta‐analysis was carried out according to the I2 statistic considering low heterogeneity when I2 < 25%, moderate heterogeneity when I2 ± 50% and high heterogeneity when I2 > 75%. A fixed‐effects model was used to estimate the pooled effects if I2 < 50%, and a random‐effects model was used when I2 ≥ ≥ 50% 22. Funnel plots of precision by logit event rate and the Begg–Mazumdar indicator were not performed because of low power where a small number of studies are included 23. The Egger's test was also not performed due to the small number of studies included 24.
Due to continuous outcome measures, standardised mean differences (SMDs) were calculated as this is more generalisable 25. If this was not possible, a mean difference was calculated. If the mean difference was lacking, a comparison was made based on the outcome measurements after the intervention.
A sensitivity analysis was performed on the following outcomes: muscle strength, muscle mass and gait speed, to investigate the impact of age (<75 years vs >75 years) and study design (only RCTs) on the effect of exercise intervention in older adults with sarcopenia. To perform the sensitivity analysis, the primary meta‐analysis was repeated, leaving out studies which did not specifically measure the specified factors. Analysis was only carried out when specified factors were measured in two or more studies.
A significance level of 5% was used for all statistical procedures. Analyses were carried out by LV and checked by WH using RevMan.
Results
Descriptive
The searches retrieved 1512 titles without duplicates. After screening titles and abstracts, 32 potential full texts were evaluated. A total of 26 studies were excluded, and six original studies were included in the review (Fig. 1) 26-31. The six included full‐text studies, dating from 2011 to 2016, enrolled a total of 480 participants of both sexes with the mean age ranging between 67.1 ± 5.3 and 81.1 ± 3.7 years. In total, 154 participants received an exercise intervention. Ninety were women, 25 were men. Two articles did not specify the gender in the exercise group 27, 31. Five of the six studies (83%) 26-30 were RCTs, and one study was a quasi‐experimental intervention study (17%) 31. The results of the data extraction are shown in Table 1.
| Author (year) | Study participants | No. | Study design | Study participants exercise group | Duration | Control group intervention | Outcome measures (measurement) |
|---|---|---|---|---|---|---|---|
| Bellomo et al. (2013) 28 | Italy | 40 | RCT | 10 participants | 12 weeks | Received information bulletins with information about the study Maintain daily habits concerning diet, social relations and physical activity |
|
| Older men with sarcopenia, aged 70 ± 5.2 years | |||||||
| Kim et al. (2012) 30 | Japan | 155 | RCT | 39 participants aged 79.0 ± 2.9 years | Three months | A health education class once a month for three months. No specific instructions on diet or physical activity were given |
|
| Older women with sarcopenia, aged 78.8 ± 2.7 years | |||||||
| Kim et al. (2013) 29 | Japan | 128 | RCT | 32 participants aged 79.6 ± 4.2 years | Three months | A health education class once a month for three months. No specific instructions on diet or physical activity were given |
|
| Older women with sarcopenia, aged ≥ 75 years | |||||||
| Maruya et al. (2016) 26 | Japan | 52 | RCT | 34 participants aged 69.2 ± 5.6 years | Six months | Maintain usual daily activities and exercise |
|
| Older people with (pre)sarcopenia, aged ≥ 60 years | |||||||
| Shahar et al. (2013) 31 | Malaysia | 65 | Quasi‐experimental intervention study | 19 participants aged 69.74 ± 5.46 years | Three months | A relaxation exercise program every two weeks for three months |
|
| Older people with sarcopenia, aged 67.1 ± 5.3 years | |||||||
| Wei et al. (2016) 27 | China | 40 | RCT | 20 participants aged 75 ± 6 years | 12 weeks | Maintain normal lifestyle and physical activity level |
|
| Older people with sarcopenia, aged ≥ 65 years |
- 10MWT, 10‐metre walk test; 5STS, five times sit‐to‐stand test; 6MWT, six‐minute walk test; 8FUGT, 8‐foot up and go test; ACT, arm curl test; BIA, bioelectrical impedance analysis; BST, back scratch test; CSRT, chair sit‐and‐reach test; CST, chair stand test; EQOLQ, Euro‐Quality of Life Questionnaire; GLFS‐25, 25‐question Geriatric Locomotive Functional Scale; Hz, hertz; KES, knee‐extension strength; OLS, one‐leg stance; RCT, randomised controlled trial; TUG, timed up and go; WBVT, whole‐body vibration training.
The total duration of the exercise training programs varied from three to six months, with a median of three months (IQR = 3). The number of exercise training sessions per week ranged from two to three, with a median of two times per week (IQR = 0.5), and the mean duration of the sessions was 46.5 minutes (SD = 23.4), ranging from 6 to 60 minutes. With regard to the type of resistance‐type exercise training performed, four studies performed whole‐body exercise training, including gait training, balance training and strength training. Maruya et al. 26 used an active home‐based exercise program, and Wei et al. 27 used whole‐body vibration training (WBVT). The control groups consisted mostly of health education 29, 30 or maintaining usual daily activities 26-28. All but one study performed group sessions 27-30, and the sixth study consisted of individual home exercises 26. Four studies performed supervised group sessions 27, 29-31, one study did not report the supervision 28, and one study had unsupervised home exercises 26. For a more detailed description of the used interventions, see Table 2.
| Give a detailed description of | Bellomo et al. (2013) | Kim et al. (2012) | Kim et al. (2013) | Maruya et al. (2016) | Shahar et al. (2013) | Wei et al. (2016) |
|---|---|---|---|---|---|---|
| The type of exercise equipment |
|
|
|
Not reported | TheraBand® exercise bands; Hygenic Corporation, OH, USA | Whole‐body vibration (WBV) machine, FitVibe Excel; GymnaUniphy NV, Bilzen, Belgium |
| The qualifications, teaching/supervising expertise and/or training undertaken by the exercise instructor | Not reported | Resistance bands: The exercises were supervised by the principal investigator, exercise instructor and assistant trainers. No qualifications or training undertaken was reported | Resistance bands: The exercises were supervised by the principal investigator, exercise instructor and assistant trainers. No qualifications or training undertaken was reported | In the intervention group, physical therapists provided a guidebook to participants | Sessions were conducted by trained exercise instructors and supervised by the researchers. No qualifications or training undertaken was reported | Sessions were supervised by a researcher. No qualifications or training undertaken was reported |
| Whether exercises are performed individually or in a group | Group sessions | Group sessions | Group sessions | Individual home exercises | Facilitated group sessions | Group sessions |
| Whether exercises are supervised or unsupervised and how they are delivered | Not reported | The exercises were supervised by the principal investigator, exercise instructor and assistant trainers. No delivery methods were described | The exercises were supervised by the principal investigator, exercise instructor and assistant trainers. No delivery methods were described | Unsupervised home exercises, provided by a guidebook |
|
The exercises were supervised by a researcher. No delivery methods were described |
| How adherence to exercise is measured and reported | Not reported | Not reported | Not reported | Reviewing participants’ daily training calendars assessed adherence to the exercise program | Not reported | Not reported |
| Motivation strategies | Not reported | Not reported | Not reported | Not reported | Not reported | Not reported |
| The decision rule(s) for determining exercise progression | The load of the resistance training per exercise was decided on the percentage of the FMT | When exercises were properly executed without significant fatigue or loss of proper execution, the resistance was increased. The principal investigator, exercise instructor and assistant trainers assessed everyone's ability to increase intensity. No detailed description of decision rules was reported | Resistance was increased on a group basis when participants could execute each exercise without loss of proper execution. Everyone's ability to increase intensity was assessed by the principal investigator, the exercise instructor and assistant trainers. No detailed description of decision rules was reported | Not reported | The intensity of the resistance exercise was adjusted according to individual performance by changing the colour of the elastic band, which had different tension levels. No detailed description of decision rules was reported | Not reported |
| How the exercise program was progressed |
The subjects performed 1 set of 15 repetitions with a load equal to 30% of FMT as specific warm‐up
|
The strengthening exercises were performed in a progressive sequence from seated to standing positions. The progressive resistance was provided using resistance bands or ankle weights. Intensity was maintained at approximately 12–14 on the Borg Rating of Perceived Exertion Scale | The strengthening exercises were performed in a progressive sequence from seated to standing positions. The progressive resistance was provided using resistance bands or ankle weights. Participants were initially instructed to complete up to one 8‐repetition set of each type of exercise, which gradually increased to 10 repetitions, and up to two sets. Intensity was maintained at approximately 12 to 14 on the Borg Rating of Perceived Exertion Scale | Not reported | Not reported | A vertical mode of WBV was used |
| Each exercise to enable replication | Not reported | There was a detailed description of most exercises | Not reported | There was a detailed description of each exercise | There was a detailed description of most exercises | There was a detailed description of the exercise |
| Any home program component | There was no home program component | There was no home program component | There was no home program component | It was a home program | Participants were encouraged to do exercise at home | There was no home program component |
| Whether there are any non‐exercise components | There were no non‐exercise components | There were no non‐exercise components | There were no non‐exercise components | There were no non‐exercise components | There were no non‐exercise components | There were no non‐exercise components |
| The type/number of adverse events that occurred | Not reported | Not reported | Not reported | Not reported | Not reported | Not reported |
| The setting in which the exercises are performed | Not reported | Tokyo Metropolitan Institute of Gerontology | Tokyo Metropolitan Institute of Gerontology | At home | Local community centre | A sports training laboratory of The Hong Kong Polytechnic University |
| The exercise intervention including, but not limited to, number of exercise repetitions/sets/sessions, session duration and intervention/program duration |
Resistance training: At the beginning of each training session, the subject carried out a warming‐up on a stationary bicycle, pedalling for 10 minutes at an intensity equal to 60% of HRmax, and performed stretching exercises for the muscles of the lower limbs. The subject performed a resistance training for lower limbs using two isoinertial exercises, leg press and leg extension
|
|
|
|
|
|
| Whether the exercises are generic (one size fits all) or tailored to the individual | The exercises were generic | The exercises were generic | The exercises were generic | The exercises were generic | The exercises were generic | The exercises were generic |
| How exercises are tailored to the individual | The load of the resistance training per exercise was decided on the percentage of the FMT | Not reported | Not reported | Not reported | The intensity was adjusted according to individual performance. No detailed description | Not reported |
| The decision rule for determining the starting level at which people commence an exercise program | The load of the resistance training per exercise was decided on the percentage of the FMT | Not reported | Not reported | Not reported | Not reported | A vertical mode of WBV was used |
| How adherence or fidelity to the exercise intervention is assessed | Not reported | Not reported | Not reported | Not reported | Not reported | Not reported |
| The extent to which the intervention was as planned | Not reported | Not reported | Not reported | Not reported | Not reported | Not reported |
- FMT, maximum theoretical force.
Methodological quality
Mean methodological quality of the included studies was 5.5 out of 10 (ranging from 0 to 10) on the PEDro scale, consisting of 11 criteria (Table 3). The item randomisation, similar baseline and interpretability of the methodological quality assessment were fulfilled by most included studies. Allocation concealment and assessor blinding was fulfilled by ≥50% of the included studies (Table 3). The inter‐rater reliability score was 0.83, with Cohen's kappa indicating substantial agreement between the raters.
| Author (Year) | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | Score/10 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Bellomo (2013) | − | + | − | − | − | − | + | + | − | − | + | 4 |
| Kim (2012) | + | + | + | + | − | − | + | + | − | + | + | 7 |
| Kim (2013) | + | + | + | + | − | − | + | + | − | + | + | 7 |
| Maruya (2016) | + | + | − | + | − | − | − | − | − | + | + | 4 |
| Shahar (2013) | + | − | − | + | − | − | − | + | − | + | + | 4 |
| Wei (2016) | + | + | + | + | − | − | − | + | + | + | + | 7 |
- 1. Eligibility criteria were specified. 2. Subjects were randomly allocated to groups (in a crossover study, subjects were randomly allocated an order in which treatments were received). 3. Allocation was concealed. 4. The groups were similar at baseline regarding the most important prognostic indicators. 5. There was blinding of all subjects. 6. There was blinding of all therapists who administered the therapy. 7. There was blinding of all assessors who measured at least one key outcome. 8. Measures of at least one key outcome were obtained from more than 85% of the subjects initially allocated to groups. 9. All subjects for whom outcome measures were available received the treatment or control condition as allocated or, where this was not the case, data for at least one key outcome were analysed by ‘intention to treat’. 10. The results of between‐group statistical comparisons are reported for at least one key outcome. 11. The study provides both point measures and measures of variability for at least one key outcome. Points are only awarded when a criterion is clearly satisfied. For the full Physiotherapy Evidence Database scale, see Appendix S2 (Supporting information).
According to the GRADE system used to assess the quality of evidence, the six studies in this review provide very low‐quality evidence that exercise interventions could be a potential strategy for reversing or slowing the sarcopenic process. The evidence was downgraded from high to very low quality because of limitations in study design, inconsistency among studies (i.e. heterogeneity ≥50% assessed by I2 in muscle strength) and lack of generalisability.
Outcome measures
Functional outcome measures
Functional outcome measurements were reported in all six studies 26-31. Muscle strength was measured in all studies; however, different strength measurements were used. Three studies 26, 29, 31 used grip strength, and five studies 26-30 used isometric knee‐extension strength. Four studies 26, 27, 29, 30 also measured gait speed, and two studies 27, 29 reported the timed up and go (TUG). The pooled outcomes of the meta‐analysis are shown in Figure 2. SMDs and 95% confidence intervals (CIs) around these SMDs are reported in Table 4.

| Bellomo et al. (2013) | Kim et al. (2012) | Kim et al. (2013) | Maruya et al. (2016) | Shahar et al. (2013) | Wei et al. (2016) | |
|---|---|---|---|---|---|---|
| Muscle mass, kg | — | d = 0.37 | d = 0.21 | — | d = −0.31 | — |
| 95% CI = −0.08 to 0.82 | 95% CI = −0.28 to 0.70 | 95% CI = −0.98 to 0.36 | ||||
| v = 0.05 | v = 0.06 | v = 0.12 | ||||
| Appendicular muscle mass, kg | — | d = 0.43 | d = 0.24 | — | — | — |
| 95% CI = −0.02 to 0.88 | 95% CI = −0.25 to 0.73 | |||||
| v = 0.05 | v = 0.06 | |||||
| Leg muscle mass, kg | — | d = 0.45 | d = 0.21 | — | — | — |
| 95% CI = −0.00 to 0.90 | 95% CI = −0.28 to 0.70 | |||||
| v = 0.05 | v = 0.06 | |||||
| Body fat, % | — | — | — | d = 0.38 | d = −0.63 | — |
| 95% CI = −0.28 to 1.03 | 95% CI = −1.31 to 0.05 | |||||
| v = 0.11 | v = 0.12 | |||||
| Grip strength, kg | — | — | d = 0.37 | d = −0.45 | d = 0.07 | — |
| 95% CI = −0.12 to 0.87 | 95% CI = −1.11 to 0.20 | 95% CI = −0.60 to 0.73 | ||||
| v = 0.06 | v = 0.11 | v = 0.12 | ||||
| Isometric knee‐extension strength, Nm | d = 2.76 | d = 0.54 | d = 0.54 | d = 0.33 | — | d = 0.46 |
| 95% CI = 1.53 to 3.98 | 95% CI = 0.09 to 0.99 | 95% CI = −0.04 to 1.04 | 95% CI = −0.32 to 0.99 | 95% CI = −0.16 to 1.09 | ||
| v = 0.39 | v = 0.05 | v = 0.06 | v = 0.11 | v = 0.10 | ||
| Usual gait speed, m/s | — | d = 1.22 | d = 0.39 | d = 0.21 | — | — |
| 95% CI = 0.73 to 1.70 | 95% CI = −0.10 to 0.89 | 95% CI = −0.44 to 0.86 | ||||
| v = 0.06 | v = 0.06 | v = 0.11 | ||||
| Timed up and go | — | — | d = −1.05 | — | — | d = −0.41 |
| 95% CI = −1.58 to −0.53 | 95% CI = −1.04 to 0.21 | |||||
| v = 0.07 | v = 0.10 |
Muscle strength
Muscle strength improved with exercise training in five of six studies. Bellomo et al. 28 reported a highly significant increase in bilateral isometric knee‐extension strength in the experimental group and no significant changes in the control group. Kim et al. 30 reported a small but not significant improvement in knee‐extension strength in the intervention group, whereas a statistically significant (P = 0.02) decrease in muscle strength was observed in their control group. Kim et al. 29 reported a non‐significant improvement in handgrip strength and knee‐extension strength in the intervention group. Maruya et al. 26 reported a statistically significant improvement in handgrip strength and knee‐extension strength in the exercise group. Shahar et al. 31 showed a decrease in handgrip strength during their intervention. Wei et al. 27 reported a significant improvement of the knee‐extension strength in the intervention group.
The meta‐analysis showed that handgrip strength did not significantly differ (Z = 0.17, CI −2.36 to 2.80, P = 0.87) between the experimental groups and the control groups (Figure 2). However, knee‐extension strength (adjusted and not adjusted for weight) did significantly differ, in favour of the experimental group (Z = 2.52, CI 0.03 to 0.26, P = 0.01; Z = 2.81, CI 3.74 to 21.03, P = 0.005, respectively).
Functional outcomes
Kim et al. 30 reported a significant improvement of the gait speed in the exercise group. Maruya et al. 26 reported a non‐statistically significant decrease in gait speed, whereas Wei et al. 27 reported a significant improvement of gait speed in their intervention groups. Both studies that measured TUG 27, 29 showed an improvement on this measurement in the exercise group.
Gait speed did not significantly improve, in the pooled meta‐analysis, in the intervention groups, in comparison with the control groups (Z = 1.84, CI −0.01 to 0.24, P = 0.07) (Figure 2). This is in contrast to the TUG, which showed a significant improvement (Z = 4.33, CI −2.43 to −0.91, P < 0.0001) in the meta‐analysis, in comparison with the control groups (Figure 2).
Body composition
Body composition was measured in four of six studies 26, 29-31. Three studies 29-31 measured the effect of exercise training on muscle mass. Two studies also measured the effect of exercise training on body fat. In all four studies that measured body composition, a bioelectrical impedance analysis was used. Bellomo et al. 28 and Wei et al. 27 did not measure body composition. The pooled outcomes of the meta‐analysis are shown in Figure 2. SMDs and 95% CIs around these SMDs are reported in Table 4.
Muscle mass
A significant improvement in leg muscle mass and a non‐significant improvement in muscle mass and appendicular muscle mass were reported in the exercise group of Kim et al. 30. Kim et al. 29 reported a small decrease in overall muscle mass and appendicular muscle mass, but a slight increase in leg muscle mass in the intervention group. Shahar et al. 31 found a trend of improved muscle mass, but no significant changes (P > 0.05) in muscle mass after the 12‐week intervention.
A pooled comparison showed that exercise therapy did not significantly improve whole‐body muscle mass (Z = 1.49, CI −0.19 to 1.37, P = 0.14) (Figure 2). It did, however, significantly improve appendicular muscle mass and leg muscle mass (Z = 2.11, CI 0.03 to 0.87, P = 0.04; Z = 2.08, CI 0.02 to 0.68, P = 0.04, respectively).
Body fat
Maruya et al. 26 reported non‐significant decreases in percentage body fat in both the intervention and control groups. Shahar et al. 31 reported a significant (P < 0.05) decrease in percentage body fat in their intervention group. In comparison with the placebo groups, exercise therapy did not significantly improve percentage body fat (Z = 0.31, CI −9.19 to 6.69, P = 0.76) (Figure 2).
Sensitivity analysis
The sensitivity analysis of the effect of exercise interventions on knee‐extension strength (not adjusted for weight) showed a significant improvement when only studies with a population with a mean age of ≥75 and when only RCT studies were included (Z = 2.51, CI 1.60 to 12.97, P = 0.01). Muscle mass as well as gait speed showed a tendency towards significance when only studies with a population with a mean age of ≥75 were included (Z = 1.80, CI −0.06 to 1.55, P = 0.07; Z = 1.74, CI −0.02 to 0.29, P = 0.08, respectively). Other outcomes of the sensitivity analysis remained non‐significant.
Discussion
The effects of exercise interventions in older adults with sarcopenia were explored in this systematic review and meta‐analysis. It appears that exercise interventions do significantly improve some, but not all, aspects of strength, functional outcomes and muscle mass. The studies were equivalent for the participants (older adults with sarcopenia) and most of the primary outcome measures (e.g. body composition, strength, gait speed), even though there was diversity between the studies in terms of the participants’ age as well as type of exercise training (resistance exercise training, WBVT and home‐based exercise).
The results of this systematic review are in line with those of another review published in 2014 32 that reported that resistance training elicited muscle hypertrophy in very old‐aged people, although the magnitude was very small 32. They also reported that muscle strength improved in these very old‐aged participants 32. The results from the current systematic review imply that exercise training interventions in older adults with sarcopenia can be effective in improving muscle mass and function. However, more research is needed to confirm these preliminary findings, as the number of studies was small, with variable quality and heterogeneous methods.
The moderate effects found in this systematic review might be due to the relatively short duration of the interventions (12 weeks). While this may be long enough to elicit measurable changes in muscle function, it may not be long enough for hypertrophy or other changes in body composition to occur, particularly in older adults with sarcopenia. A randomised controlled trial conducted in 2015 reported that despite the interindividual variability in the adaptive response, all older participants improved in body composition and functional outcome measures after resistance exercise training 33. However, the duration of the exercise training was an important factor. There were individuals who demonstrated little to no improvement after 12 weeks, but a substantial improvement after 24 weeks of training 33. This systematic review identified five of six studies with an intervention of 12 weeks, which might have resulted in similar limited results. The moderate effects found might also be due to the relative lack of progressive overload in the exercise interventions. However, no definitive conclusions can be drawn from this due to the poor reporting of the interventions performed. The lack of detailed description makes it impossible to reflect on the progressive resistance training approaches used. Other limitations of the included articles include not using a dual‐energy X‐ray absorptiometry scan to measure body composition and shortcomings due to low statistical power observed in the trials.
A recently published study reported that exercise habits during middle age are associated with increased muscle strength and lower prevalence of sarcopenia in older age 34. Exercise habits in middle age may be a protective factor against sarcopenia in older age 34. The sensitivity analysis of the current review shows a tendency towards significance on both functional and body composition outcomes, when only studies with a population with a mean age of >75 were included. More long‐term longitudinal studies are necessary to confirm an effective role of exercise interventions in preventing and reversing sarcopenia. In addition, longer follow‐up is needed to assess the long‐term effects of exercise interventions on functional outcomes and body composition.
The aim of this review was to determine the effect of exercise interventions in older adults with sarcopenia; therefore, only exercise training interventions were included. Other published articles have also included nutrition interventions and/or supplement interventions within their exercise groups, as complex interventions are considered one of the mainstays of interventions for sarcopenia. However, a systematic review from the EWGSOP in 2014 reported that this evidence is also based on short‐term studies and large clinical trials are still lacking 5. Interventions that evaluated the combined effects of exercise and nutrition/supplementation have suggested a potential additive effect, but the existing evidence is inconsistent. Their overall message was that there are many gaps in the evidence and that further research is needed 35.
A limitation of the current review is the different assessments of function and different training strategies that most likely explain the high heterogeneity of muscle strength and balance outcomes in the meta‐analysis. This illustrates that despite the growing literature and the consensus of different associations, such as the EWGSOP, Asian Working Group for Sarcopenia and the National Institutes of Health, there is still ambiguity in the measurement and treatment of older adults with sarcopenia.
Due to the lack of published data in the articles, no mean standard deviations could be calculated. Attempts were undertaken to obtain the required information, such as standard deviation of the difference or correlation coefficients, but overall this information was lacking and could not be included in this systematic review. Although using a comparison of final measurements to perform a meta‐analysis is an alternative method presented 24, we must take into account that this might have a negative effect on the statistical power of the meta‐analysis performed. More research with better statistical analysis descriptions is needed to conduct a meta‐analysis based on mean differences. Also, a sensitivity analysis on exercise prescriptions would have been preferable, comparing interventions that are likely to improve muscle mass and strength and interventions that are focused on a general health benefit or improving cardiovascular function. However, due to the lack of description in the articles, this was not possible.
The very few RCTs conducted using exercise interventions in older adults who were identified with sarcopenia is another limitation. This may have been partly caused by the choice to specify the search from 2006 onwards. Also, the lack of consensus on the definition of sarcopenia, and the associated difficulty to include participants, may have held back larger RCTs from taking place, resulting in the low numbers of included studies.
The training effects from the included RCTs were inconsistent. This resulted in a low quality of evidence and emphasises that more homogeneous research is needed to confirm these preliminary results. Despite these limitations, the extensive search and the reasonable methodological quality strengthen the findings in this systematic review.
The burden of sarcopenia is extensive and it is still increasing with population ageing, thus prevention and effective interventions are important. Cost‐effectiveness of interventions is also important. Currently, there are no cost‐effective exercise intervention trials that focus on older adults with sarcopenia. However, in previous studies with different study populations, exercise interventions were shown to be more cost‐effective compared with dietary interventions 36.
Conclusion
More research is needed to determine the effects of exercise interventions in adults who already have sarcopenia. Additional research is needed, specifically focused on the timing of exercise interventions, stage of sarcopenia and the progressive resistance training approaches. Also, researchers should report a detailed description of the interventions used, to enable the possibility to better reflect the used training load and the associated training effects. Although research on exercise interventions in older people with sarcopenia is advancing, many questions still remain.
Acknowledgement
The authors declare no conflicts of interest.




